[Talk-us] Special issues in LA remap

Alan Mintz Alan_Mintz+OSM at Earthlink.Net
Wed Jun 13 16:31:06 BST 2012


At 2012-06-07 16:39, Mike N wrote:
>  Using way id 13292685 : http://www.openstreetmap.org/browse/way/13292685 
> , I looked at the edits.  In this case, some nodes were moved, presumably 
> to align with the aerial imagery.  I couldn't find any GPS tracks in the 
> immediate area to confirm the aerial alignment.   The node I looked at 
> was moved at least 10 meters.
>...
>    A fast preemptive remap for nodes set for redaction-move would be to 
> just move the node to align with current Bing aerial.   Note that the 
> Bing aerial alignment might not be perfect - try to find some GPS tracks 
> in the area to see how close it is.  There should be some tracks on the 
> nearest interstate.

This is a non-issue in southern CA. I've done enough calibration against 
high-accuracy benchmarks in the area to believe that Bing imagery is 
uniformly within 2m of true. OTOH, GPS tracks, especially from an antenna 
inside a  moving vehicle, are much less accurate than that.


>   So, the redaction bot will not completely destroy those areas.  If no 
> one has touched the road alignment since his work, it will just revert 
> back to the original TIGER upload coordinates.  All that is required 
> after the redaction bot will be to nudge the nodes to the location where 
> they belong.

10m is a big shift. If the errors were that big in this area, I would bet 
that if you looked at the original TIGER05 import, it would look like a 
toddlers scrawl compared to the look of the map now. I've seen untouched 
areas before that were that bad, so I know what it looks like. "All that is 
required" will be a huge amount of work to re-align thousands of ways. It 
seems far easier to look at them, compare against the imagery, and 
pronounce them accurate, and I'd like to get a real legal opinion from 
someone involved with the licensing on whether the latter approach would be 
sufficient. Can we do that?

--
Alan Mintz <Alan_Mintz+OSM at Earthlink.net>




More information about the Talk-us mailing list